Fetal Surgery for Congenital Heart Disease

Updated: Jun 26, 2017
  • Author: Anita J Moon-Grady, MD, FACC, FAAP, FASE; Chief Editor: Hanmin Lee, MD  more...
  • Print


There is no viable animal model that recapitulates the anatomy and physiology observed in obstructive right- and left-side cardiac lesions in the human. However, advances in ultrasound technology have enabled earlier and more precise diagnosis of human fetal cardiac lesions; this technology has also enabled the field of congenital heart disease to gain greater understanding of the unique fetal hemodynamics and the mechanisms involved in the evolution of cardiac disease in utero.

Given the now substantial body of knowledge regarding the fetal physiology and natural history of these lesions in utero and the success of balloon aortic and pulmonary valvuloplasty in preventing or reversing new-onset ventricular dysfunction postnatally in infants, there is a theoretical rationale for intervention to relieve valvar stenosis in fetal life. There is also a physiologic and practical argument for enlarging the atrial septal communication in fetuses with restrictive interatrial septum and left atrial hypertension, who will otherwise have prenatal pulmonary abnormalities and postnatal instability due to a small or nonexistent interatrial communication.

Intervention for cardiac defects fits into existing fetal care structures [1] and is an extension of services and clinical research protocols related to fetal treatment. However, it is still largely considered an adjunct to optimal postnatal management and does not obviate the need for further intervention after birth.

Owing to the complications of any intrauterine intervention, investing in innovative fetal treatment can be considered a useful exercise only when it is made feasible in terms of procedural ease and risks.

Experimental studies on open fetal cardiac surgery began in the 1980s in animal models to study the physiologic and pathologic mechanisms of extracorporeal circulatory bypass. [2] Although initial research showed promise of reproducibility in technique, the fetoplacental response to bypass, characterized by an end result of fetal hypoxia and demise, deterred complete success.

Factors contributing to placental dysfunction following a time on bypass appeared to be multiple, with reports of cytokine-mediated injury evidenced by the increase in prostanoids, vasoactive leukocytes, complement, and consequent endothelial dysfunction. [3, 4, 5, 6] Maternal placental dysfunction also followed owing to an increase in vascular resistance, and a relative state of fetal hypoxia ensued as a consequence of the impairment in fetal blood flow. [7, 8, 9] As a result, fetal open cardiac surgery was largely abandoned.

The earliest reported human fetal cardiac therapy of any kind was in 1975 and involved maternal-fetal transplacental administration of a beta blocker in the setting of fetal ventricular tachyarrhythmia. [10] The first open in-utero cardiac procedure was reported a decade later, in 1986, with a pacemaker placement for complete heart block. [11]

The concept of performing balloon valvuloplasty in fetuses with stenotic heart valves followed the successful introduction of neonatal balloon valvuloplasty in the 1980s, with the first reported case performed in a fetus with aortic stenosis in 1989. [12]

Modification of early techniques has since allowed percutaneous access for catheter-based interventions and has met with technical success, not only in fetal aortic valve stenosis but also in pulmonic valve atresia and stenosis and in hypoplastic left heart syndrome (HLHS) with an intact or highly restrictive atrial septum. [13, 14]

Reports of clinical developments from the largest series of these procedures confirm technical success and reproducibility. [15, 16]  They also suggest that successful fetal procedures lead to improvement in functional chamber development and myocardial function while the fetus is still in utero.

Other centers have reproduced the initial successes, [17, 18] as well as provided evidence that fetal valvuloplasty in conditions of atretic or stenotic valves of the aorta and pulmonary artery can facilitate the chance of biventricular circulation after birth, whereas septoplasty for intact or severely restrictive interatrial septum may improve postnatal stability and chances of survival after initial palliative surgery. [19, 20]

It is also possible that the intrauterine environment is naturally conducive to wound healing and regeneration at a cellular level. [21] However, much effort must be applied to appropriate planning and selection of candidates (both maternal and fetal) to minimize the obvious complications and risks associated with in-utero interventions.

Furthermore, local oversight should dictate whether, on procedure-based grounds, the proposed fetal intervention constitutes human subjects research, innovative therapy, or clinical care, as well as ensure that the appropriate counseling and consent procedures are followed. Information from an international registry of cases suggests that although the majority consider the actual fetal procedure to be clinical care, there is some regional variability in consent requirements and research ethics interpretations. [16]



Indications for fetal cardiac intervention have included the following:

In May 2014, the American Heart Association (AHA) issued a scientific statement on the diagnosis and treatment of fetal cardiac disease, [22] in which it was stated that fetal catheter intervention may be considered in the following (class IIb recommendation; level of evidence, B/C):

  • Fetuses with aortic stenosis with antegrade flow and evolving HLHS
  • Fetuses with aortic stenosis, severe mitral regurgitation, and restrictive atrial septum
  • Fetuses with HLHS with a severely restrictive or intact atrial septum
  • Fetuses with PA-IVS

Patient selection

Appropriate patient selection involves evaluation not only of the fetal cardiac defect but also of the maternofetal aspect and the pregnancy as a whole. Given that these defects have a spectrum of severity and usually progress during gestation, the ideal time to treat them is before any additional insult is incurred. Therefore, the ideal gestational age is when the patient meets criteria and as soon as possible.

These defects also have significant differences in their disease progression. Aortic stenosis clearly has a degree of severity at which there is a point of no return in terms of evolution to HLHS. In contrast, in pulmonary atresia with intact ventricular septum, the right ventricle may have more capacity to remodel (based on observations in postnatal patients) and likely does not incur as much myocardial damage in utero, for unknown reasons.

The ideal gestational age for treatment is therefore between 18 and 30 weeks’ gestation for aortic stenosis, with 22-30 weeks preferred for pulmonary valve procedures and up to 34 weeks preferred for an atrial septal procedure in the presence of left heart obstruction.

Initially, the fetal cardiac defect and the potential for successful alteration of the fetal anatomy and physiology in such a way as to achieve the goal of the intervention (viable left or right ventricle in aortic stenosis or pulmonary atresia; left atrial decompression in restrictive or intact atrial septum) should be critically evaluated by an experienced physician team, including pediatric/fetal cardiologists.

Specific suggested criteria for aortic valvuloplasty in the fetus were published in 2011 [23]  and were reevaluated in 2017 at the same center [24] ; however, criteria for intervention in pulmonary atresia and atrial septal restriction are less clear, and none exist for fetal pacing (though a clinical trial with a prototype device is in the planning stages).

Whether invasive confirmation of a normal karyotype in the fetus is required is controversial, but such confirmation should be considered desirable. In regions where noninvasive maternal serum testing is available, this should be performed and the fetus found to be euploid before intervention.  There is equipoise regarding whether this is a reasonable substitute for invasive testing; therefore, this becomes a region- or center-specific issue. [25]

Additionally, a detailed ultrasound examination should be performed to exclude other fetal anomalies and to define placental location, cervical length, and uterine or adnexal masses that could influence the safety and technical aspects of the procedure.

A thorough evaluation of the maternal health and obstetric history should be undertaken by an experienced maternofetal medicine specialist to rule out any absolute contraindications to maternal anesthesia or the procedure itself and to disclose any potential relative contraindications that would have to be discussed prior to proceeding. When feasible, the proposed procedure should also be discussed with the referring obstetrician/perinatologist who will be caring for the patient after the procedure.

All patient counseling is preferably done with the partner or support person(s) present, and every effort should be made to ensure that there is an understanding of the issues regarding the procedure’s safety and efficacy for both the pregnant woman and the fetus. The clinicians are required to provide patients with clear verbal and (if required by local oversight agencies) written information and with counseling and support both before and after the procedure.

Patients who do not demonstrate a clear understanding of the potential risks and uncertain long-term benefit of these procedures should be considered incapable of providing informed consent and should not be considered candidates for fetal intervention.



Contraindications for fetal cardiac intervention include the following:

  • Significant preexisting maternal disease or obstetric comorbidity that would place the fetus or mother at higher risk, including anesthesia and the invasive procedure itself; a body mass index (BMI) exceeding 35 (relative contraindication); other relative contraindications (which should be weighed against potential benefit) include, but are not limited to, maternal communicable diseases such as HIV infection, uncontrolled or pregestational diabetes, history of cervical incompetence, and hematologic disorders that affect coagulation
  • Significant extracardiac pathology in the fetus (including significant chromosomal abnormality and structural abnormalities other than the cardiac lesion)
  • Multiple gestation (relative)
  • Inability of the mother to provide informed consent

Technical Considerations

Fetal aortic valvuloplasty in severe stenosis

Severe obstructive lesions, including HLHS, can evolve from a simple semilunar valve obstruction during the gestational period (see the videos below). [26, 27, 28, 29] This observation initially prompted experimental attempts at fetal balloon valvuloplasty of the aortic and pulmonary valves in the 1980s and 1990s, which showed that technical success was possible but did not show improved outcome. [12, 30]

Fetal aortic stenosis at 20 weeks' gestation. Left ventricle is dilated, and ventricular function is poor, because of severe obstruction of aortic valve.
Same fetus as in previous video, now at 34 weeks' gestation, with evolving hypoplastic left heart syndrome due to aortic stenosis present earlier in gestation. Left ventricle is now small, is echo-bright, and shows no systolic contraction.

The first reported successful series of per-ventricular fetal aortic valvuloplasty to halt the evolution of HLHS in fetuses with aortic stenosis from a single center was published by Tworetzky et al in 2004. [15] Of the 20 pregnancies in which this procedure was performed, 14 (70%) cases were technically successful, with 21% of fetuses subsequently achieving biventricular circulation after birth.

Interestingly, this group also reported the natural history in a control group of affected pregnancies wherein intervention was offered but declined. [15] In this observational cohort, all fetuses developed hypoplasia of the left side requiring single-ventricle palliation by term.

With growing experience in fetal cardiac catheter-based interventions and modifications of the technique, a larger series of treated pregnancies (including the original cases) was reported 5 years later. [31] Of the 68 fetuses who underwent aortic balloon valvuloplasty, 15 (22%) went on to have a successful biventricular outcome as neonates. In the last 50 cases, the success rate was 31% (27% as neonates, with an additional 4% converted to a biventricular circulation at a later stage in their palliation).

Another group from Linz, Austria, reported that as many as 43% of fetal aortic stenosis cases treated with successful prenatal intervention led to biventricular circulation after birth, with 67% of live-born technical successes being biventricular. [18] Other factors (eg, older gestational age at the time of intervention and, perhaps, less severe left ventricular dysfunction at the time of the procedure) may have influenced the higher success rates seen in this series in comparison with the initial Boston series.

Subsequent reports revealed a better understanding of fetal aortic valvuloplasty in general, with success rates comparable to those reported by the Linz group, demonstrating the importance of optimal patient selection and procedural timing, as well as overall advances in postnatal care. [17, 32]

In a study by Freud et al that evaluated 100 fetuses who had undergone fetal aortic valvuloplasty for severe midgestation aortic stenosis with evolving HLHS (median follow-up, 5.4 years), 38 of the 88 who were born alive had a biventricular circulation, either from birth or after initial univentricular palliation. [33]

Larger dimensions of left heart structures and higher left ventricular pressure have also been retrospectively recognized as predictors of successful fetal aortic valvuloplasty and eventual biventricular circulation, [31] whereas moderate-to-severe endocardial fibroelastosis at the time of procedure (thought to be due to long-standing obstruction in some cases as opposed to a primary myocardial destructive process) is associated with lack of response despite technically successful valvuloplasty. [34]

In a 2017 reevaluation of a large series of aortic stenosis fetuses over a 15-year period, [24]  technical success and biventricular outcome were assessed in patients with physiologic success at birth. Independent predictors of biventricular outcome included the following:

  • Higher left ventricular pressure
  • Larger ascending aorta
  • Better ventricular diastolic function (longer/biphasic inflow pattern)
  • Higher left ventricular long-axis Z-score

The authors were able to assign high, intermediate, and low likelihood of biventricular outcome to patients on the basis of their retrospective data.

Postnatal evaluation confirms that the newer aggressive valvar reparative strategies and resection of endocardial fibroelastosis have resulted in sufficient left ventricular rehabilitation to yield a functional biventricular circulation. [32, 35] The longer-term outcome of these patients, however, is still to be determined.

Although the potential for prevention of left heart hypoplasia with such invasive strategies is exciting, there are important considerations that limit routine application. Patient selection criteria and timing of intervention are important factors toward achieving success. [26, 31]

Although Doppler demonstration of flow reversal across the foramen ovale and retrograde perfusion of the transverse aortic arch have been identified as markers of severe fetal aortic stenosis potentially associated with progressive left heart hypoplasia (see the images below), [36] all hypoplastic changes may not necessarily sequence from the stenosis as the primary lesion. Conversely, not all will benefit from fetal intervention or even need it; many may be amenable to postnatal left ventricular rehabilitation without fetal intervention. [37]

Fetus with severe aortic stenosis and left-to-righ Fetus with severe aortic stenosis and left-to-right blood flow by color Doppler through foramen ovale (arrow).
Sagittal image of fetus with severe aortic stenosi Sagittal image of fetus with severe aortic stenosis. Aortic arch is filling retrograde in systole, as evidenced by color Doppler, with normal antegrade flow (blue) in ductal arch and retrograde flow (red) in aortic arch.

Appropriate patient selection is critical. It is now appreciated that left ventricular myocardial disease associated with endocardial fibroelastosis can also result from infection, maternal autoantibodies, and other less common etiologies, which may be difficult to distinguish from aortic stenosis as the primary cause (see the images below). [38, 28, 27] .

Endocardial fibroelastosis: gross anatomy. Image c Endocardial fibroelastosis: gross anatomy. Image courtesy of Phil Ursell, MD, Department of Pathology, University of California, San Francisco, School of Medicine.
Endocardial fibroelastosis: histopathology. Image Endocardial fibroelastosis: histopathology. Image courtesy of Phil Ursell, MD, Department of Pathology, University of California, San Francisco, School of Medicine.

Considerable further effort must be applied to modeling myocardial structure and function in healthy, treated, and untreated groups so as to help define and elucidate any distinguishing features that may improve patient selection. Clinical functional outcomes appear to be influenced not only by the degree of hypoplasia alone but also by the degree of mitral valve disease, aortic hypoplasia, and the degree of endocardial fibroelastosis of the left ventricle.

Overall, it is clear from present studies that in-utero balloon valvuloplasty as an isolated procedure has not obviated the need for additional postnatal intervention, including repeat aortic valvuloplasty, repair of coarctation, resection of endocardial fibroelastosis, mitral valvuloplasty, and temporary left atrial decompression procedures. [30, 35, 18, 20, 32, 31]

Restrictive or intact atrial septum

It is well established that HLHS confers a worse prognosis when it is associated with a severely restrictive or intact atrial septum, resulting in an early mortality of 50-90%. [39] Prenatal signs of left atrial hypertension include a dilated chamber with septal bulging (see the video below). [19]

Fetal hypoplastic left heart syndrome with intact interatrial septum. Note thickened, bulging interatrial septum with left atrial hypertension evidenced by pulmonary vein dilation.

The increase in left atrial pressure resulting in pulmonary vascular changes may not be reversible after birth, resulting in severe pulmonary edema and cyanosis and demonstrated as pulmonary venous flow abnormalities. [40, 41, 42] Fetal intervention for restrictive atrial septum was proposed to decompress the left atrium as a means of alleviating the development of these pulmonary vascular complications. [39, 19, 20]

Atrial septoplasty is performed by placing a needle through the restricted septum into the fetal right atrium or through the left atrial free wall. In one series, of 19 successful cases after fetal atrial septoplasty, 12 required additional procedures in the immediate postnatal period and seven remained stable until the Norwood procedure could be performed. Two fetuses died in utero. However, more than half of the cohort survived through the operation, compared with only 10-20% in whom fetal atrial septoplasty had not been performed. [20, 43]

As early results revealed a gestational age–related exponential diminution in the size of these defects, it has been suggested that improved longer-term patency might be achieved by using coronary artery stents.

Although intervention holds potential promise for this small subset of affected fetuses, recognition of severe left atrial hypertension due to restrictive atrial septum before birth and its timing of evolution appear to be critical factors for success. The authors and other researchers have demonstrated that pulmonary venous Doppler evaluation shows reduced forward flow in early diastole and increasing reversed flow in atrial systole with increasing left atrial hypertension. [40, 42, 44]

The authors’ conclusion is that altered pulmonary venous Doppler patterns with only forward flow in ventricular systole and reversed flow in atrial systole are more likely to be associated with critical cyanosis after delivery, making the performance of atrial septoplasty a true emergency requirement, whereas the presence of even a small amount of forward flow in early diastole may identify the fetus in whom intervention may not be necessary for several hours to days, if at all.

However, the longer-term repercussions of this chronic in-utero left atrial hypertension remains unknown; data suggest worse outcomes in these fetuses, even if urgent postnatal decompression is unnecessary. [44]

This raises an interesting question: Would fetal intervention then be justified even in patients who were expected to be stable in the neonatal period? Clearly, more investigation into this area and the development of animal models of left atrial hypertension may help shed light on the optimal timing of intervention and the potential for reversal of vascular changes that may improve longer-term outlook for these single-ventricle patients.

Fetal pulmonary atresia with evolving hypoplastic right heart

Fetal intervention for severe right ventricular outflow obstruction via transthoracic per-ventricular balloon pulmonary valvuloplasty has also been attempted. The role of intervention is to promote right heart functional development and to improve the possibility of postnatal biventricular circulation (see the video below).

Fetal pulmonary atresia with intact ventricular septum. Right ventricle is small and hypertrophied at 27 weeks, with tricuspid valve size 3 standard deviations below normal for gestational age. This would represent borderline candidate for fetal procedure, as success with postnatal therapy alone is likely.

Isolated case reports and one series have described this procedure. [17, 45, 46, 47] In a series of 10 affected fetuses, a group from Boston reported technical success in 60% of fetuses along with subsequent in-utero growth of right heart structures, 40% of whom were able to achieve biventricular circulation after birth. [17]

Identification of potential candidates is generally based on the risk of progression to univentricular circulation if left untreated, though this risk is less well defined for fetuses with right heart obstruction than for those with aortic valve obstruction.

Reviews of the evolution of severe pulmonary outflow tract obstruction have shown that a preprocedure tricuspid valve annulus dimension of more than 3 standard deviations below gestational age–based normal values and diminutive size of the pulmonary valve and the outflow [47, 48, 49, 46] are more likely to be associated with evolution of right heart hypoplasia sufficient to preclude biventricular circulation at birth despite aggressive postnatal intervention. This should guide future attempts at fetal pulmonary valvuloplasty.

Therefore, although technical success is possible for fetal aortic and pulmonary valvuloplasty and atrial septoplasty, further work is needed before these techniques are widely adopted, in part because of the rarity of the lesions for which intervention would be warranted.

For instance, of all cases of HLHS encountered in one large referral center over several years, only 1.6% involved a dilated left heart and suspected critical aortic stenosis at initial prenatal diagnosis. [50] Fewer than one quarter of infants with left-side hypoplasia have been observed to have some degree of clinically important atrial septal restriction, and many of these infants may have sufficient decompression in the preoperative period. [43]

Ideally, randomized multicenter trials that include long-term follow-up of patients who have undergone these procedures will provide the most definitive conclusions as to efficacy, safety, and superiority over current palliative pathways that include only postnatal interventions. Given the rarity of the fetal disease, however, issues related to the cost and feasibility of such trials may continue to be prohibitive in the current medical climate.

Fetal complete heart block

Complete heart block in the fetus can be a life-threatening emergency that may present without warning. Diagnosis relies on fetal M-mode and Doppler ultrasonography to relate atrial and ventricular mechanical events. Heart block in the fetus may be associated with structural heart disease, [51] may result from maternal autoimmune disease, or may be idiopathic. [52]

In most cases associated with a structurally normal heart, maternal autoantibodies are present in the maternal serum. The mothers may not show clinically apparent symptoms of lupus or Sjögren syndrome; however, in both, transplacental passage of the antibodies can potentially cause inflammation and fibrosis at the atrioventricular (AV) node in particular.

Although some success has been reported with administration of corticosteroids to the mother of a fetus with first- or second-degree AV block, pharmacologically effective options are available for treatment of third-degree or complete heart block, and fetuses with low intrinsic ventricular rates are at risk for significant morbidity and mortality. Administration of beta-mimetics to increase fetal heart rate have somewhat improved fetal survival. [53, 54, 55]

A few isolated investigators over the past two decades have attempted treatment to increase ventricular rates in the face of hydrops and imminent demise of the fetus in the form of implantation of fetal pacemakers. [56, 57] The results appeared mixed, with some reporting initial good results, with complete resolution of hydrops and achievement of normal heart rates.

Theoretically, this approach should provide relief in the gestational period wherein the insult is the most severe, prolonging pregnancy and promoting feasibility of early postnatal permanent pacemaker implantation. However, to date, no survivors following institution of fetal pacing have been reported.

Previous approaches have relied on the placement of an epicardial pacing lead on the fetal heart with an extrauterine pulse generator implanted in the maternal abdomen, with failure attributed to lead dislodgment due to fetal movement. A prototype device that is delivered via a transcatheter technique may be more promising and is currently in preclinical testing. [58]  Owing to the lack of information regarding successful performance of this procedure, this approach is not discussed further in this article.  

Technical limitations or failures of fetal interventions

Although it is evident that technical success is possible for fetal aortic and pulmonary valvuloplasty and atrial septoplasty, universal adoption of in-utero treatment for all fetuses is not appropriate. There are geographical and economic considerations owing to the complexity of planning involved. The level of care both in the obstetric environment and for neonatal cardiac intensive care may not be available at all institutions.

Currently, the practice of neonatal cardiac surgery is confined to a few institutions that have invested a great deal of time and effort in research, development, and refinement of technical expertise in this area. In addition, much remains to be learned about the benefits and potential adverse effects of intervention.

The use of these procedures is also complicated by technical issues related to fetal positioning, stabilization, and the need for high-fidelity continuous imaging by specially trained personnel with experience in ultrasound guidance of invasive procedures. There are equipment-related limitations, including imaging artifacts caused by the materials used to manufacture the needles, wires, and catheters, exacerbated by the diminutive sizes of the cardiac structures being imaged.

Because of the relatively recent development of these procedures, there is also a paucity of data regarding their long-term outcomes; reliable outcome data are vital if such interventions are ever to become routine in clinical practice.

Ideally, for all forms of intrauterine cardiac intervention, randomized multicenter trials with long-term follow-up would provide the most definitive conclusions as to their efficacy, safety, and superiority over current palliative pathways that include only postnatal interventions. To date, limited clinical data have not shown significant improvements in terms of functional or neurodevelopmental status [33, 59] ; however, the numbers are small.

Without long-term data, it is not possible to address the ethical dilemma posed by weighing the risks of fetal intervention against the current surgical success rate of 95% that some centers have achieved in postnatal Norwood stabilization procedures. [60, 61, 62] The intrinsic risks to both mother and fetus associated with in-utero procedures may thus be too high to allow advocacy of routine fetal cardiac intervention until more information about long-term functional and neurodevelopmental outcome is available.

Future directions

Generally, nonsurgical cardiac therapies (including pharmacologic treatments) for the management of fetal tachyarrhythmias and bradyarrhythmias are better established than in-utero interventions and much more amenable to universal adoption, in that they avoid the risks uterine interventions invariably carry. Nevertheless, it is indisputable not only that fetal cardiac intervention is possible but also that giant strides have already been made toward establishing technical expertise in this area, even though routine clinical application outside a few highly specialized centers is still far off.

The therapeutic utility and staying power of fetal cardiac intervention depend on a host of critical issues. Clinical and research efforts must focus on enhancing our understanding of the natural history of these conditions, as well as our ability to predict the myocardial response to invasive fetal intervention, so as to improve patient selection and increase the risk-benefit ratio beyond its current state. Education with a focus on earlier diagnosis of congenital heart disease and timely appropriate referral to centers capable of performing the interventions will be key.

In view of the relative rarity of these fetal congenital heart conditions, more collaborative effort [23, 16] is necessary in order to improve our procedural techniques and to achieve an acceptable safety level for both the mother and fetus.

It is also essential to continue efforts to improve equipment technology for diagnostic imaging and in the procedure room. One potentially important advance may come in the form of sophisticated robot-guided fetal cardiac intervention. [63] Fetal therapy for congenital heart conditions is an area that holds great promise in the management of complex cardiac disease.